<?xml version="1.0" encoding="UTF-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
  <title>GitIntel Blog</title>
  <subtitle>Research and insights on AI-generated code in open source.</subtitle>
  <link href="https://gitintel.com/blog" rel="alternate"/>
  <link href="https://gitintel.com/atom.xml" rel="self"/>
  <id>https://gitintel.com/blog</id>
  <updated>2026-04-11T07:30:05.544Z</updated>
  <entry>
    <title><![CDATA[Two Years of AI Progress. Security Rate: Still 55%.]]></title>
    <summary><![CDATA[Veracode tested 100+ AI models. 45% introduce known security flaws — identical to 2024. The security rate hasn't moved in two years.]]></summary>
    <link href="https://gitintel.com/blog/ai-code-security-flat-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/ai-code-security-flat-2026</id>
    <published>2026-04-11T00:00:00.000Z</published>
    <updated>2026-04-11T00:00:00.000Z</updated>
    <category term="Security"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[AI Tools Leave Invisible Fingerprints in Your Git History. Researchers Can ID Them at 97% Accuracy.]]></title>
    <summary><![CDATA[A January 2026 arXiv paper analyzed 33,580 PRs from 5 AI coding agents (Codex, Copilot, Devin, Cursor, Claude Code) and identified which tool wrote the code at 97.2% F1-score — even when developers didn't disclose it.]]></summary>
    <link href="https://gitintel.com/blog/ai-agent-fingerprints-git-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/ai-agent-fingerprints-git-2026</id>
    <published>2026-03-30T00:00:00.000Z</published>
    <updated>2026-03-30T00:00:00.000Z</updated>
    <category term="Research"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[AI Makes Developers 19% Slower. They Think They're 20% Faster.]]></title>
    <summary><![CDATA[METR's RCT: experienced developers are 19% slower with AI while believing they're 20% faster — a 39-point gap. BCG's March 2026 HBR study: 14% suffer 'AI brain fry,' error rates 39% higher. The productivity trap has peer-reviewed data.]]></summary>
    <link href="https://gitintel.com/blog/ai-productivity-trap-burnout-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/ai-productivity-trap-burnout-2026</id>
    <published>2026-03-30T00:00:00.000Z</published>
    <updated>2026-03-30T00:00:00.000Z</updated>
    <category term="Research"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[AI Writes Your JSON. 1 in 4 Times, It's Wrong. ICLR 2026 Has the Data.]]></title>
    <summary><![CDATA[University of Waterloo's StructEval tested 12 AI models on 2,035 structured output tasks. GPT-4o tops out at 76%. Text→Mermaid collapses to 18.9%. The format-specific failure crisis is now peer-reviewed.]]></summary>
    <link href="https://gitintel.com/blog/ai-structured-output-failure-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/ai-structured-output-failure-2026</id>
    <published>2026-03-30T00:00:00.000Z</published>
    <updated>2026-03-30T00:00:00.000Z</updated>
    <category term="Research"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[Dario Amodei Said 90% of Code Would Be AI-Written. The Deadline Passed 6 Months Ago.]]></title>
    <summary><![CDATA[Amodei predicted 90% AI-written code by September 2025. GitHub reports 51% AI-assisted. GitIntel scans average 5.8%. The data behind the most debated prediction in software — and why the metric was unfalsifiable to begin with.]]></summary>
    <link href="https://gitintel.com/blog/amodei-90-percent-code-myth-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/amodei-90-percent-code-myth-2026</id>
    <published>2026-03-30T00:00:00.000Z</published>
    <updated>2026-03-30T00:00:00.000Z</updated>
    <category term="Research"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[Your AI Coding Agent Gets Dumber the Longer It Works. Here's the Proof.]]></title>
    <summary><![CDATA[All 18 frontier models tested by Chroma Research degrade with context length. A January 2026 arXiv paper found models miss their advertised window by 99%+. Some start failing at 1,000 tokens. Here's what that means for AI-generated code.]]></summary>
    <link href="https://gitintel.com/blog/context-rot-coding-agents-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/context-rot-coding-agents-2026</id>
    <published>2026-03-30T00:00:00.000Z</published>
    <updated>2026-03-30T00:00:00.000Z</updated>
    <category term="Research"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[Shadow AI: 65% of Enterprise AI Tools Are Unapproved. Source Code Is the #1 Data They're Leaking.]]></title>
    <summary><![CDATA[Harmonic Security: source code is 30% of all sensitive data leaked via shadow AI — ahead of legal docs and M&A data. IBM: shadow AI breaches cost $670K more. 90% of enterprises use AI. 37% have governance policies. EU AI Act enforcement: August 2026.]]></summary>
    <link href="https://gitintel.com/blog/shadow-ai-enterprise-code-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/shadow-ai-enterprise-code-2026</id>
    <published>2026-03-30T00:00:00.000Z</published>
    <updated>2026-03-30T00:00:00.000Z</updated>
    <category term="Security"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[The AI Agent Cost Crisis Nobody's Talking About]]></title>
    <summary><![CDATA[Developers are 19% slower with AI tools (METR study). 45% of AI code has OWASP vulnerabilities. Your team spends $15-25K/month on AI coding tools. Nobody's tracking the ROI.]]></summary>
    <link href="https://gitintel.com/blog/ai-agent-cost-crisis" rel="alternate"/>
    <id>https://gitintel.com/blog/ai-agent-cost-crisis</id>
    <published>2026-03-29T00:00:00.000Z</published>
    <updated>2026-03-29T00:00:00.000Z</updated>
    <category term="Research"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[AI Wrote Your Code. Now Your Churn Is Up 41%.]]></title>
    <summary><![CDATA[GitClear analyzed 153M lines of code and found AI tools have driven churn up 41%, duplication up 4x, and refactoring to a 10-year low. The quality crisis is measurable and growing.]]></summary>
    <link href="https://gitintel.com/blog/ai-code-churn-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/ai-code-churn-2026</id>
    <published>2026-03-29T00:00:00.000Z</published>
    <updated>2026-03-29T00:00:00.000Z</updated>
    <category term="Research"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[Your AI-Generated Code Can't Be Copyrighted. But It Can Infringe Someone Else's.]]></title>
    <summary><![CDATA[Supreme Court March 2026: pure AI code has no copyright protection under US law. But it can still infringe yours. 51% of GitHub commits are AI-assisted with zero documentation. EU AI Act enforcement hits August 2026.]]></summary>
    <link href="https://gitintel.com/blog/ai-code-copyright-crisis-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/ai-code-copyright-crisis-2026</id>
    <published>2026-03-29T00:00:00.000Z</published>
    <updated>2026-03-29T00:00:00.000Z</updated>
    <category term="Legal"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[Developers Are Opening 98% More PRs With AI. Their Reviews Got 91% Longer.]]></title>
    <summary><![CDATA[CodeRabbit reviewed 13M PRs: AI-coauthored PRs have 1.7x more issues. Incidents per PR up 23.5% YoY. PR size up 154%. The review bottleneck data nobody's talking about.]]></summary>
    <link href="https://gitintel.com/blog/ai-code-review-crisis-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/ai-code-review-crisis-2026</id>
    <published>2026-03-29T00:00:00.000Z</published>
    <updated>2026-03-29T00:00:00.000Z</updated>
    <category term="Research"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[AI Code Gets Smarter. It Doesn't Get Safer.]]></title>
    <summary><![CDATA[Veracode Spring 2026: syntax pass rates hit 95%, security pass rates flat at 45-55% since 2023. 35 CVEs from AI code in March alone. XSS fails 86% of the time. The divergence is now official.]]></summary>
    <link href="https://gitintel.com/blog/ai-code-security-gap-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/ai-code-security-gap-2026</id>
    <published>2026-03-29T00:00:00.000Z</published>
    <updated>2026-03-29T00:00:00.000Z</updated>
    <category term="Security"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[AI Writes Your Code. Nobody Understands It. Here's the Data.]]></title>
    <summary><![CDATA[88% of developers say AI worsens technical debt. Maintenance costs hit 4x by year two. Addy Osmani named it Comprehension Debt — the gap between how much code exists and how much your team actually understands.]]></summary>
    <link href="https://gitintel.com/blog/ai-comprehension-debt-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/ai-comprehension-debt-2026</id>
    <published>2026-03-29T00:00:00.000Z</published>
    <updated>2026-03-29T00:00:00.000Z</updated>
    <category term="Research"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[AI Tools Hallucinate Package Names. Attackers Are Registering Them.]]></title>
    <summary><![CDATA[LLMs suggest non-existent npm/PyPI packages in ~5-20% of coding tasks. Attackers register those names with malicious payloads. Slopsquatting is the newest supply chain attack — and 51% AI-assisted commits means the attack surface is now enormous.]]></summary>
    <link href="https://gitintel.com/blog/ai-dependency-hallucination-supply-chain" rel="alternate"/>
    <id>https://gitintel.com/blog/ai-dependency-hallucination-supply-chain</id>
    <published>2026-03-29T00:00:00.000Z</published>
    <updated>2026-03-29T00:00:00.000Z</updated>
    <category term="Security"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[AI PRs Are Flooding Open Source. Maintainers Are Drowning.]]></title>
    <summary><![CDATA[AI PRs generate 1.7x more issues than human ones (10.83 vs 6.45 findings). PRs up 20% YoY. cURL killed its bug bounty. LLVM added AI-disclosure rules. The review bottleneck is here.]]></summary>
    <link href="https://gitintel.com/blog/ai-pr-flood-open-source-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/ai-pr-flood-open-source-2026</id>
    <published>2026-03-29T00:00:00.000Z</published>
    <updated>2026-03-29T00:00:00.000Z</updated>
    <category term="Research"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[AI Coded It. You Can't Debug It. Anthropic Has the Data.]]></title>
    <summary><![CDATA[Anthropic's Jan 2026 RCT: AI users scored 17 points lower on post-task skills tests than controls. 50% vs 67%. No measurable speed gain. The skill atrophy is documented and growing.]]></summary>
    <link href="https://gitintel.com/blog/ai-skill-atrophy-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/ai-skill-atrophy-2026</id>
    <published>2026-03-29T00:00:00.000Z</published>
    <updated>2026-03-29T00:00:00.000Z</updated>
    <category term="Research"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[AI Writes Your Tests. Coverage Goes Up 40%. Bugs Stay.]]></title>
    <summary><![CDATA[AI tools boost test coverage 40% and cut test-writing time 50%. But defect rates haven't moved. 1.75x more logic errors in AI code, AI-generated tests that validate the wrong behavior, and the 'coverage theater' pattern showing up in production postmortems.]]></summary>
    <link href="https://gitintel.com/blog/ai-test-coverage-trap-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/ai-test-coverage-trap-2026</id>
    <published>2026-03-29T00:00:00.000Z</published>
    <updated>2026-03-29T00:00:00.000Z</updated>
    <category term="Research"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[73% of Developers Use AI Every Day. 97% Don't Fully Trust It. Here's the Data.]]></title>
    <summary><![CDATA[15,000 developers surveyed. 73% use AI coding tools daily — up from 18% in 2024. Only 3% highly trust the output. METR measured the actual productivity gap. The verification problem nobody's solved.]]></summary>
    <link href="https://gitintel.com/blog/ai-trust-gap-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/ai-trust-gap-2026</id>
    <published>2026-03-29T00:00:00.000Z</published>
    <updated>2026-03-29T00:00:00.000Z</updated>
    <category term="Research"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[Cursor Hit $2B Revenue. Copilot Has 4.7M Users. Neither Leaves a Trace in Your Git History.]]></title>
    <summary><![CDATA[51% of GitHub commits are AI-assisted. Cursor has $2B ARR. GitHub Copilot has 4.7M subscribers. None of them leave attribution in your git history — and that's a compliance crisis.]]></summary>
    <link href="https://gitintel.com/blog/cursor-copilot-git-attribution-crisis" rel="alternate"/>
    <id>https://gitintel.com/blog/cursor-copilot-git-attribution-crisis</id>
    <published>2026-03-29T00:00:00.000Z</published>
    <updated>2026-03-29T00:00:00.000Z</updated>
    <category term="Research"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[What's Trending on GitHub: March 2026]]></title>
    <summary><![CDATA[OpenClaw 263K stars. Ollama 162K. n8n 150K. Dify 130K. GitHub has 4.3M AI repos (+178% YoY). Three patterns dominate: local-first, open-source infrastructure, and the agent plugin explosion.]]></summary>
    <link href="https://gitintel.com/blog/github-trending-march-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/github-trending-march-2026</id>
    <published>2026-03-29T00:00:00.000Z</published>
    <updated>2026-03-29T00:00:00.000Z</updated>
    <category term="Trends"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[How last30days-skill Built a 14K-Star Research Engine in One File]]></title>
    <summary><![CDATA[14,449 stars. 10 platforms. One SKILL.md file. We break down the architecture of the most popular Claude Code skill on GitHub — query detection, source tiering, scoring, and 10 patterns you can steal.]]></summary>
    <link href="https://gitintel.com/blog/last30days-skill-analysis" rel="alternate"/>
    <id>https://gitintel.com/blog/last30days-skill-analysis</id>
    <published>2026-03-29T00:00:00.000Z</published>
    <updated>2026-03-29T00:00:00.000Z</updated>
    <category term="Architecture"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[The MCP Ecosystem Just Hit 8,600 Servers. Here's What That Means.]]></title>
    <summary><![CDATA[873% growth in nine months. 8,600+ MCP servers. Zero standardized security vetting. The protocol powering AI agents has a trust problem — and the EU AI Act deadline is four months away.]]></summary>
    <link href="https://gitintel.com/blog/mcp-ecosystem-explosion" rel="alternate"/>
    <id>https://gitintel.com/blog/mcp-ecosystem-explosion</id>
    <published>2026-03-29T00:00:00.000Z</published>
    <updated>2026-03-29T00:00:00.000Z</updated>
    <category term="Security"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[The AI Coding Benchmark War Has a Legitimacy Crisis. Here's the Data.]]></title>
    <summary><![CDATA[Claude Opus 4.5 scores 80.9% on SWE-Bench Verified and 45.9% on SWE-Bench Pro. 35-point gap. OpenAI abandoned Verified. 59.4% of its hardest problems have flawed tests. The benchmark everyone cites is broken.]]></summary>
    <link href="https://gitintel.com/blog/swe-bench-legitimacy-crisis-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/swe-bench-legitimacy-crisis-2026</id>
    <published>2026-03-29T00:00:00.000Z</published>
    <updated>2026-03-29T00:00:00.000Z</updated>
    <category term="Research"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[Vibe-Coded Apps Are Shipping With Critical Security Holes. Here's the Data.]]></title>
    <summary><![CDATA[10.3% of Lovable-generated apps had critical RLS flaws. 21% of YC W'25 companies are 91%+ AI-generated. 63% of vibe coders are non-developers. The security bill is coming due.]]></summary>
    <link href="https://gitintel.com/blog/vibe-coded-apps-security-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/vibe-coded-apps-security-2026</id>
    <published>2026-03-29T00:00:00.000Z</published>
    <updated>2026-03-29T00:00:00.000Z</updated>
    <category term="Security"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[5 Viral Repo Patterns Every AI Builder Should Know (March 2026)]]></title>
    <summary><![CDATA[We analyzed 5 repos that collectively gained 150K+ stars this month. The patterns they share are about memory, not features — subconscious hooks, temporal decay, research skills, peer messaging, and swarm orchestration.]]></summary>
    <link href="https://gitintel.com/blog/viral-repo-patterns" rel="alternate"/>
    <id>https://gitintel.com/blog/viral-repo-patterns</id>
    <published>2026-03-28T00:00:00.000Z</published>
    <updated>2026-03-28T00:00:00.000Z</updated>
    <category term="Research"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[We Scanned 13 Popular Open Source Repos for AI-Generated Code. Here's What We Found.]]></title>
    <summary><![CDATA[Deno is 41% AI-coded. Claude Code appeared in 99% of AI commits. Data from 6,500 commits across Next.js, Ruff, LangChain, Ollama, and more.]]></summary>
    <link href="https://gitintel.com/blog/ai-code-in-open-source-2026" rel="alternate"/>
    <id>https://gitintel.com/blog/ai-code-in-open-source-2026</id>
    <published>2026-03-22T00:00:00.000Z</published>
    <updated>2026-03-22T00:00:00.000Z</updated>
    <category term="Research"/>
    <author><name>GitIntel Research</name></author>
  </entry>
  <entry>
    <title><![CDATA[The Official Claude Code Telegram Plugin Has 3 Tools. Here's What's Missing.]]></title>
    <summary><![CDATA[We mapped every Telegram Bot API capability against the official plugin. 13 features missing. We built them — inline keyboards, scheduler, document handling, and more.]]></summary>
    <link href="https://gitintel.com/blog/claude-code-telegram-plugin-limitations" rel="alternate"/>
    <id>https://gitintel.com/blog/claude-code-telegram-plugin-limitations</id>
    <published>2026-03-22T00:00:00.000Z</published>
    <updated>2026-03-22T00:00:00.000Z</updated>
    <category term="Engineering"/>
    <author><name>GitIntel Research</name></author>
  </entry>
</feed>